The constructive role of models
A model is a representative device. But it is unlike either of the two main traditional classes of sign. It is unlike a natural sign in that there is no causal relation between a model and its subject, while a natural sign like smoke is related causally to the fire that it signifies. It is unlike a conventional sign, since the choice of a particular model as a representation is not wholly arbitrary. The selection of a model is based upon real resemblances and differences between the model and what it represents, and decisions as to what are the proper degrees of likeness and unlikeness can be the subject of argument. The term ‘model’ captures the main features of this mode of representation better than the somewhat similar notion of ‘picture’. Its range of significance is more suited to the role of this kind of representative device in science than that of the latter term. For example, a model may be a concrete object capable of comparable performances to the thing modelled. The behaviour itself may be the model, in which case we would quite properly want to call that behaviour a process that is a model of another process. We may want to speak of something imaginary that is not yet realized or may not be capable of concrete realization, as a model. The mode of representation in pictures, at least traditionally, is more thoroughly projective than that by which certain important categories of models are related to their subject matter. The likeness between a model and its subject matter may be no more than a formal homology. I shall use the word ‘model’ as the
most general term of this kind, reserving ‘picture’ for a certain class of models. Diagrams, using highly schematized representative devices, but following certain projective conventions between themselves and their subject matter, form the borderline case of that sub-class of models, properly called pictures.